Inter-Alkane Peaks Analysis Approach by Means of Machine Learning Algorithms for Reservoir Connectivity Assessment on The Example of a Site in Eastern Siberia

Summary The paper proposes a technique based on the areas of inter-alkane peaks and includes the calculation of the maximum possible number of normalized ratios of paeks areas, preliminary data processing, application of principal component analysis and k-means clustering algorithm. It allows not only to evaluate the reservoir connectivity, but also to determine the number of principal components required for research, helps in choosing the optimal number of groups within which the reservoirs are connected, and also offers a new approach to the selection of inter-alkane peak ratios for displaying oil samples on radar charts.